| Prestressed concrete continuous box girder Bridges are widely used as the main bridge types with single hole span ranging from 40m to 150m.In order to obtain the distribution of temperature field and stress field of box girder concrete and guide temperature control measures,finite element software is usually used to simulate the concrete of box girder.However,in the simulation calculation,the thermal parameters of concrete obtained by laboratory tests and empirical formulas are often different from the actual situation of field construction.Therefore,in order to accurately obtain the thermal parameters of box girder concrete,it is necessary to study the back analysis method of box girder concrete.The main work contents of this paper are as follows:(1)ABAQUS finite element software was used to establish the transient thermal analysis model of box girder concrete,HETVAL subroutine was used to define the heat generated by hydration heat of concrete,thermal parameters of concrete were determined by reference to empirical formula,and the temperature field and stress field of box girder concrete at the initial stage of construction were obtained.Based on the finite element model of box girder concrete,Based on the finite element model of box girder concrete,the sensitivity analysis of 9 parameters including specific heat capacity C,density ρ,thermal conductivity λ,final heat of hydration Q0,cement quantity Wc,entering temperature Tp,parameters a and b in the double exponential expression of cement hydration heat,and outdoor heat release coefficient β1 of box girder concrete box is carried out.The analysis methods include single factor sensitivity analysis and sensitivity analysis based on orthogonal test range analysis and variance analysis.The study found that the results of the three sensitivity analysis methods are basically identical.Parameters Wc,Q0,ρ,C and b are highly sensitive to the temperature peak of hydration heat of box girder concrete,while parameters Tp,β1,λ and a are relatively weak.(2)The back analysis model was established by BP neural network and differential evolution algorithm(Different Evolution,DE)optimized BP neural network.Based on uniform design and box girder concrete finite element model,130 groups of data were generated for neural network training and testing.Mean absolute percentage error(MAPE)was used to evaluate the prediction effect of neural network.The MAPE of BP neural network is less than 4%,but the relative error of some data is more than 10%.Meanwhile the MAPE of DE-BP neural network is less than 3%,and the relative error of all test samples is less than 5%.This shows that the differential evolution algorithm can effectively avoid the neural network falling into the local minimum value.Therefore,DE-BP neural network has good predictive ability and can be used to invert the thermal parameters of box girder concrete.(3)Taking Shitan Xiangjiang River Bridge as the engineering background,28 temperature measuring points were arranged before concrete pouring of box girder,the temperature change curve of box girder concrete was obtained by monitoring the temperature for 120 hours after the concrete was poured.The temperature variation curve was fitted and the expression of temperature variation with time was obtained.The temperature peaks of the characteristic points were input into the DE-BP neural network to obtain the thermal parameters,and then the temperature change curves of each point were obtained by finite element calculation.It is found that the maximum difference of peak temperature of each characteristic point is 2.05℃,and the calculated values of web and bottom plate agree well with the measured values.However,the calculated and measured values of the roof feature points are in poor agreement after the temperature decreases,which is mainly due to other environmental factors such as solar radiation and wind speed are not considered in the finite element model.Therefore,the results of back analysis using DE-BP neural network can accord with the reality. |
PDF Full Text Request